Ardea
Official journal of the Netherlands Ornithologists' Union

login


[close window] [previous abstract] [next abstract]

Tinbergen J.M. (1981) Foraging decisions in Starlings (Sturnus vulgaris L ). ARDEA 69 (1): 1-67
1) Six seasons of field work on the Dutch island Schiermonnikoog were devoted to testing the hypothesis that parent Starlings conform to the principle of caloric maximization in their choice of prey and hunting site as would be predicted by the profitability concept introduced by Royama. 2) The study is based on direct observations of intake rate of the Starlings coupled with automatic photographic records of prey delivered to the nest (Fig. 7). The latter allow subsequent measurement of prey length and hence estimate of weight of the prey. Since the parents were individually colour-marked, detailed time-budget observations were possible (Fig. 5) and in the final season search paths and prey captures were measured synchronously. 3) The menu of the nestling Starlings is dominated by only a few species in all years. These were collected from the pastureland surrounding the colony ('polder') or from the adjoining salt marsh (Fig. 10). The chief prey collected from the polder is the Leatherjacket Tipula paludosa, whereas the salt marsh provides the caterpillar Cerapteryx graminis, the beetle Telephorus fuscus and in some years other important prey (Table 4). 4) After each feeding the parent Starling must decide which prey to collect next and since Tipula and Cerapteryx occur in quite distinct habitats, it is possible for the observer to quantify parental choice in this particular dichotomy. In the course of the day the time required to collect one prey varies in a characteristic fashion, showing a clear mid-day peak in apparent availability of the Leatherjacket (Fig. 13) but the rate of delivery of prey to the nest does not follow these trends (Fig. 12). The governing influence of nestling hunger revealed by manipulations of the nestlings (see point 5) helps to explain these discrepancies particularly as nestling hunger state has an obvious and deep-seated relation to time of day. 5) Time required to collect one Cerapteryx from the salt marsh is far in excess of time needed for one Tipula from the nearby polder, and from the viewpoint of caloric return on time or energy expended by the parent it is not clear why Cerapteryx is brought at all. Field experiments on choice (either with prey fully visible, Table 6, or with 'prey concealed under flaps in feeding tables, Fig.18) confirm the preference of the parent Starling for Cerapteryx. Apparently other criteria have priority, and manipulation of the state of hunger of the brood provide a strong case that the quality of the prey (i.e. nutrient composition) is the reason for including Cerapteryx in the diet. When the level of nestling demand is not' too high, the parents persist in collecting caterpillars, but when demand is high (nestlings deprived and/or supernormal broods, see Figs. 19, 20, 21) parents revert to the easier-to-collect Leatherjacket. There is some evidence (Fig. 23) that Leatherjackets are detrimental to the health of the nestlings if fed in large quantities over long periods of time, so we are dealing here with a compromise situation. An indirect assessment of parental diet (Fig. 22) shows that the female parent reduces her own caterpillar consumption when stressed by the large broods, she then delivers all caterpillars captured to the young at the cost of a pure Leatherjacket diet for herself to ensure them of a varied diet. 6) Having decided which prey to collect, the next decision facing the parent Starling is where to land. This problem is considered for the Leatherjacket area where detailed observation of landing sites and subsequent search paths were collected using an optical rangefinder (Fig. 25). The main body of data refer to hunting by one female Starling on seven consecutive days in the 1979 season (Fig. 27). Summation of search path measurements in this period (Fig. 28) reveal a concentration of hunting in three exploitation centres, which can be contrasted with three marginal zones. Direct observation showed that the bird spent most of its time in the areas yielding the highest intake rates (Fig. 29) and an extensive sampling programme indicated that these coincide with the areas with highest prey density, allowing the conclusion that the threshold for exploitation at this time was about 70 Leatherjackets/ m2 (Fig. 29 top). By mounting a feeding table near one nest and measuring usage by the local male (Fig. 30) in relation to reward rate (manipulated by the observer) the causal link between intake rate and site choice was verified. 7) The history of exploitation of Leatherjackets, particularly the pattern of landings in time, strongly argues for the use of topographic memory. Analysis of landings in relation to the previous take-off point, depending on whether a high or a low intake rate had been experienced there, showed that landings tended to be close if intake had been high, and farther away if previous intake had been low (Fig. 32). This negative correlation between landing distance and prior intake rate persists even on the finest scale achieved (Fig. 33, data grouped in meter classes). That this fine-scale topographic memory may be overruled in some situations is apparent when switches to other prey habitats intervene (Fig. 35). Use of topographic cues on a meter by meter scale, implies at the very least, a clumping in prey distribution, a pattern which has been verified by absolute sampling at the smallest plot-size considered (25 cm2, see Table 11). 8) Is the threshold intake rate for Leatherjacket exploitation discovered for the individual Starling observed in 1979 always the same? By considering visits to 400 m2 plots in the Leatherjacket grid in relation to directly observed intake rates (Fig. 36) thresholds were found to vary from year to year. The simplest criterion for the individual bird is to rely on its accumulated experience on intake rates, and set the threshold equal to the mean for that period (Fig. 37). Laboratory tests confirm the adjustability of the rejection threshold (Fig. 38). 9) The simplest form of foraging trip consists of a flight out, collection of prey at the hunting site, and flight back to the nest with prey. Depending on the roundtrip flight time as well as the cumulative intake, Charnov & Orians' model (Fig. 40) can be used to predict the optimal load size, i.e. how many prey should the Starling collect for its nestlings in one trip. Since the fit to the empirical data is a reasonable one (Figs. 41 and 42) further analysis of the mechanisms the parent relies on in deciding on load size should be undertaken. 10) Does the parent Starling deplete the prey stock to such an extent that its intake rate suffers, hence leading to abandonment of the site? Our best data on the short-term influence of prey removal on subsequent hunting success results from experiments with small enclosures on the Leatherjacket grid where two captive birds were introduced, one after the other. The second bird always suffers a depressed intake rate (Fig. 47), an effect much stronger than can be explained by the absolute number of prey removed by the first bird. The most likely explanation for this is that the prey are differentially available for capture. This finding leads one to expect that birds will tend to abandon a site after only a small proportion of the prey stock has been removed, since even at modest levels of depletion intake rate will decline sharply. The 7-day Leatherjacket exploitation cycle watched in detail yields the estimate, that only twenty percent of the prey. were removed from the best areas by all the Starlings together (Fig. 45). It follows that prey sampling alone, even though completely accurate, is inadequate to predict impact and/or potential intake rate of the predator. The problem of differential availability and the related phenomenon of depleting the catchable fraction, i.e. creaming off the prey, is an unwelcome reality that must be faced in any field study. 11) In the final synthesis, it is argued that two facets from the Starling work deserve incorporation in general foraging theory. Discontinuities in the distribution of the prey cannot be adequately described by employing a one-level concept of patch. Instead, it is argued that patches exist at two levels (Fig. 50): At one level micro-patches can be distinguished; these are the indivisible units of prey distribution characterized by MacArthur's repeatability concept within which prey encounter is random; due to these features and their very small scale, exploitation is short-term. At the other level we have the macro-patch. This consists of a cluster of micro-patches and prey distribution is of low repeatability or even unique: the bird exploits these centres by relying on long term spatial memory. It is at this level that intake rate has decisive influence. Although it can be argued that the habitat characters causing discontinuities in prey distribution do in fact form a continuum we prefer to distinguish two discrete levels of organization because we feel these are entities in the decision structure of the predator. 12) The second facet deemed widely applicable is the hypothesis that the decision on prey species precedes the decision on where to go (Fig. 52). We argue that current profitability theory offers a reasonable basis for interpreting the decisions governing choice of site within one prey species, but we wish to emphasize that the decision on which prey species to collect next is of a higher order in the hierarchy of decisions: in the nestling situation this is influenced by the compromise between the physiological demands of the brood and the foraging limits of the parent. Although we do not yet understand how these limits are set, both a limit on hunting time as well as energy expended are liable to be involved. Although there is concrete evidence for the existence of the fine-scale spatial level in the decision hierarchy (level III corresponding with the macropatch) we have inserted an undefined abstraction at the level of a conglomeration of macropatches ('area') to cover the possibility that there is some generalization (for example in 'remembering' intake rates) above level III.


[close window] [previous abstract] [next abstract]